RQ-2026-009815 - 2x Senior Artificial Intelligence (AI) Software Developer for Referral RAG and MCP

Project Overview

The Agency is expanding the Provincial Health Services Directory (PHSD) to incorporate artificial intelligence capabilities that improve provider discovery, referral quality, and healthcare navigation across provincial digital health applications. The project will develop intelligent AI agents that leverage Retrieval Augmented Generation (RAG), Model Context Protocol (MCP), and FHIR-based healthcare data to assist healthcare providers in identifying appropriate referral destinations, validating referral information, recommending alternative providers or services, and improving the completeness and quality of referral submissions. The solution will utilize PHSD provider, practitioner, organization, healthcare service, specialty, location, and service metadata together with clinical referral context to provide intelligent recommendations while maintaining traceability to authoritative source data. AI agents will consume PHSD FHIR APIs, Smile CDR, AWS HealthLake, the Agency Master Data Management, and other provincial digital health assets to provide grounded responses rather than generative opinions. The project includes development of reusable AI services, MCP servers, semantic search capabilities, vector-based retrieval, prompt orchestration, evaluation frameworks, and secure integration with SMART on FHIR applications and provincial healthcare systems. The implementation will emphasize explainability, auditability, privacy, security, and clinical confidence while supporting future AI-enabled healthcare workflows across the Agency.

Key Responsibilities

• Design and develop AI agents supporting intelligent provider and referral destination discovery.
• Develop Retrieval Augmented Generation (RAG) pipelines utilizing PHSD healthcare data.
• Design and implement MCP servers exposing PHSD functionality to AI clients.
• Develop semantic search capabilities supporting provider, practitioner, organization, healthcare service, and specialty discovery.
• Integrate AI agents with Smile CDR, AWS HealthLake, PHSD APIs, Provider Registry, and Master Data Management services.
• Develop intelligent referral destination recommendations based on referral reason, specialty, geography, language, service availability, and organizational relationships.
• Develop AI services that validate referral completeness and identify missing referral information prior to submission.
• Implement explainable AI responses referencing authoritative PHSD records and supporting evidence.
• Develop prompt templates, orchestration workflows, tool selection strategies, and AI evaluation frameworks.
• Optimize retrieval quality, ranking, semantic search relevance, latency, and operational performance.
• Implement monitoring, evaluation, telemetry, audit logging, and continuous improvement processes for AI services.
• Produce technical documentation, AI governance documentation, operational procedures, and knowledge transfer materials.
• AI architecture, technical design documentation, and implementation plans supporting PHSD Referral Intelligence.
• AI agents capable of assisting users in identifying appropriate referral destinations using PHSD provider directory information.
• Retrieval Augmented Generation (RAG) pipelines utilizing PHSD, Smile CDR, AWS HealthLake, and other approved healthcare data sources.
• Model Context Protocol (MCP) servers exposing PHSD capabilities for AI-assisted applications.
• Semantic search implementation supporting provider, practitioner, organization, healthcare service, specialty, and location discovery.
• Vector embedding generation, indexing, retrieval optimization, and relevance tuning.
• AI services capable of validating referral destinations, identifying incomplete referral information, and recommending appropriate providers or healthcare services.
• Integration with PHSD FHIR APIs, Smile CDR, AWS HealthLake, Provider Registry, Master Data Management services, and other provincial digital health assets.
• Prompt libraries, orchestration workflows, tool definitions, and structured AI interaction patterns.
• AI evaluation framework including groundedness, retrieval quality, factual accuracy, hallucination detection, citation validation, and response quality metrics.
• Secure authentication and authorization integration supporting SMART on FHIR, OAuth2/OIDC, and role-based access controls.
• Operational dashboards, monitoring, telemetry, audit logging, and performance reporting for AI services.
• Unit testing, integration testing, AI evaluation testing, deployment automation, and CI/CD pipeline support.
• Technical documentation, operational procedures, AI governance documentation, knowledge transfer materials, and production implementation support.

Qualifications & Requirements

REQUIRED

• Must be able to work onsite as needed
• Enterprise AI application development using modern LLM frameworks and agent architectures.
• Retrieval Augmented Generation (RAG) architecture, semantic retrieval, vector embeddings, and knowledge grounding.
• Model Context Protocol (MCP) server development, tool integration, and agent orchestration.
• Prompt engineering, tool calling, structured outputs, evaluation frameworks, and AI workflow design.
• Integration of AI solutions with FHIR R4/R4B, SMART on FHIR, REST APIs, and healthcare interoperability standards.
• Java Spring Boot and/or Python backend development supporting AI services.
• Amazon Bedrock, OpenSearch vector search, embeddings, inference APIs, and enterprise AI deployment.
• Healthcare provider directory, referral workflows, practitioner, organization, healthcare service, and location data modelling.
• Secure API development using OAuth2, OpenID Connect, JWT, and enterprise authentication patterns.
• Cloud-native AWS development, monitoring, logging, CI/CD, and operational support.
• Strong analytical, problem solving, communication, and technical documentation skills.
• Experience evaluating AI accuracy, hallucination reduction, grounding, and response quality.

NICE TO HAVE

• Experience developing AI solutions for healthcare provider directory, referral, scheduling, or care navigation systems.
• Experience implementing healthcare knowledge graphs or semantic healthcare search.
• Experience with Amazon Bedrock Agents or other enterprise AI orchestration platforms.
• Experience implementing MCP-based enterprise integrations.
• Experience developing AI evaluation frameworks including groundedness, factuality, retrieval accuracy, and hallucination detection.
• Experience implementing FHIR terminology services, SNOMED CT, LOINC, and healthcare ontologies.
• Experience with clinician-facing AI assistants and decision support tools.
• Experience with Responsible AI, privacy-preserving AI, explainability, and AI governance.
• Experience working with Agile delivery teams building enterprise healthcare AI platforms.

EVALUATION CRITERIA

• Enterprise AI development experience, including Retrieval Augmented Generation (RAG), Large Language Models (LLMs), semantic search, prompt engineering, AI agent development, Model Context Protocol (MCP), tool orchestration, and AI evaluation frameworks. - 25 points
• Healthcare interoperability experience, including FHIR R4/R4B, SMART on FHIR, Smile CDR, AWS HealthLake, REST APIs, Provider Registry, Master Data Management, healthcare terminology, and healthcare data integration. - 25 points
• Cloud application development experience, including Java/Spring Boot and/or Python, AWS cloud services, Amazon Bedrock, OpenSearch vector search, secure API development, CI/CD, monitoring, and scalable cloud-native architectures. - 25 points
• Experience working with digital health assets Hospital Report Manager (HRM), Enterprise Master Data Management, Consent Management Service, provincial identity services, and integration with Epic, Oracle Health (Cerner), MEDITECH, or other healthcare information systems. - 25 points

Key Skills & Competencies

AI agents, AI evaluation, AI governance, AI workflow design, Amazon Bedrock, Amazon Bedrock Agents, AWS AI, Agile delivery, agent architectures, agent orchestration, agent workflows, care navigation, CI/CD, Cloud-native AWS development, decision support tools, embeddings, enterprise AI deployment, enterprise AI orchestration, enterprise AI solutions, enterprise authentication, enterprise healthcare AI platforms, explainability, factuality, FHIR R4/R4B, FHIR terminology services, grounding, groundedness, hallucination detection, hallucination reduction, healthcare AI, healthcare interoperability standards, healthcare knowledge graphs, healthcare ontologies, healthcare provider directory, healthcare service, Health Services Directory, inference APIs, Java Spring Boot, JWT, knowledge grounding, Large Language Models, location data modelling, logging, LOINC, MCP, modern LLM frameworks, monitoring, OAuth2, OpenID Connect, OpenSearch, operational support, organization, practitioner, privacy-preserving AI, prompt engineering, prompt orchestration, Python, RAG, referral systems, referral workflows, response quality, Responsible AI, REST APIs, retrieval accuracy, role-based authorization, scheduling, Secure API development, semantic healthcare search, semantic retrieval, semantic search, SMART on FHIR, SNOMED CT, structured healthcare data, structured outputs, tool calling, tool integration, vector databases, vector embeddings, vector search